* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
Download Means for minmizing pulse reflections in linear delay lines loaded
Ground (electricity) wikipedia , lookup
Power engineering wikipedia , lookup
Electromagnetic compatibility wikipedia , lookup
Immunity-aware programming wikipedia , lookup
Variable-frequency drive wikipedia , lookup
Mercury-arc valve wikipedia , lookup
Time-to-digital converter wikipedia , lookup
Electrical ballast wikipedia , lookup
Chirp compression wikipedia , lookup
Power inverter wikipedia , lookup
History of electric power transmission wikipedia , lookup
Three-phase electric power wikipedia , lookup
Pulse-width modulation wikipedia , lookup
Electrical substation wikipedia , lookup
Transformer types wikipedia , lookup
Distribution management system wikipedia , lookup
Power electronics wikipedia , lookup
Nominal impedance wikipedia , lookup
Schmitt trigger wikipedia , lookup
Resistive opto-isolator wikipedia , lookup
Two-port network wikipedia , lookup
Voltage regulator wikipedia , lookup
Switched-mode power supply wikipedia , lookup
Voltage optimisation wikipedia , lookup
Power MOSFET wikipedia , lookup
Zobel network wikipedia , lookup
Stray voltage wikipedia , lookup
Current source wikipedia , lookup
Surge protector wikipedia , lookup
Alternating current wikipedia , lookup
Network analysis (electrical circuits) wikipedia , lookup
Buck converter wikipedia , lookup
Dec. 13, 1955 R. J. SLUTZ 2,727,143 MEANS FOR MINIMIZING PULSE REFLECTIONS IN LINEAR DELAY LINES LOADED WITH A NONLINEAR LOAD Filed Aug. 30, 1951 2 Sheets-Sheet 1 +62 Ii INCIDENT\ §Ei PULSE FRONT / E° Z'a 7"", “70 LEVEL To OFCUEEENT REQ. BVGATE i E,- \ / T EEFLECTED WAVE 13 T2 INVENTOR. 15 17445 Ralph IS/uz‘z T4 =1 BY M f . M ATTORNEY Dec. 13. 1955 R. J. SLUTZ 2,727,143 MEANS FOR MINIMIZING PULSE REFLECTIONS IN LINEAR DELAY LINES LOADED WITH A NONLINEAR LOAD Filed Aug. 30, 1951 2 Sheets-Sheet 2 +62 V —81/ :71 I ____. w» . F59. 5'62 "651’ +627! ‘81/ I 21 “15 +21! F. 4127 . F925 17 INVENTOR. BYRalp/v 1.571412 11]"TORA/EV 2,727,143 United States Patent 0 Patented Dec. 13,: 1955 2 1 2,727,143 a certain magnitude will be attenuated or shorted to a much greater degree than pulses below that magnitude, so that excess pulse voltage above that needed for circuit utilization Will be rendered ineffective to produce harm MEANS FOR MINIMIZING PULSE REFLECTIONS 1N LINEAR DELAY LINES LOADED WITH A NONLINEAR LOAD ful re?ections. Ralph J. Slutz, Kensington, Md.,'assignor to the United the termination of a delay line, said shunt being returned States of America as represented by the Secretary of Commerce ,’ It is a speci?c object to provide a nonlinear shunt for to a low voltage source to prevent the production of re ?ected pulses of su?iciently high amplitude to cause trouble in the utilization circuits. 10 Application August so, 1951, Serial Nb. 244,447 Another object is to provide a termination'for a delay 4 Claims. (01. 250-27) line transmitting pulses which presents to any excess voltage of such pulses an apparent impedance not greater (Granted under Title 35,U. S. Code (1952), sec. 266) than the characteristic impedance of the utilization line The invention described herein may be manufactured 15 whereby re?ected pulses will be either absent or else of opposite sign from the incident pulses. and used by or for the Government of the United States for governmental purposes without the payment to me of any royalty thereon in accordance with the provisions of the Act of March 3, 1883, as amended (45 Stat. 467; 35 U. S. C. 45). - This invention is concerned with minimizingpulse re ?ections in linear delay lines loaded with a nonlinear load. This is a serious problem in the operation of elec tronic digital computers of the type exempli?ed by the The speci?c nature of the invention, as well as other objects and advantages thereof, will clearly appear from a description of a preferred embodiment, as shown in the accompanying drawings in which: Figure 1 is a schematic circuit diagram of a basic gate circuit. Figure 2 is a simpli?ed diagram showing those com ponents of the circuit of Figure l which are essential to a computer known as SEAC (Standards Eastern Automatic 25 consideration of the invention. Figure 2a shows graphically the changes with respect to Computer) which is in operation at the National Bureau time which occur in the incident and re?ected voltage of Standards. and current in the circuit of Figure 2. SEAC uses time-sequenced electronic pulses of approxi Figure 3a, b, c, and d represent respectively the essential mately 50 percent duty factor, i. e., one-half microsecond in duration and spaced approximately one-half micro 30 load component of the basic circuit of Figure 1 and vari ous modi?cations applied thereto in accordance with my second apart to represent digits and orders. Such pulses invention. are electrically switched through various circuits for the Figure 4a, b, c, and a‘ represent graphically the im purpose of performing desired computations at very high pedance conditions in Figure 3a, b, c, and a’, respectively. speed. Absolute accuracy in transmitting the pulses is, Referring to Figure 1, an incoming pulse on the pri-. of course, essential to the proper operation of such a de 35 mary of transformer 1 is emitted from the secondary and vice. Synchronization of pulses is maintained by using a suitably delayed by arti?cial delay line 2, which may be so-called “clock pulse” generator as a pulse source and of any known type, and has the characteristics shown in referring all pulses in the system to said source. The the drawing, the particular value of impedance being pulses are often delayed somewhat in transition through a utilization circuit and in order to synchronize them with 40 selected, together with the impedance of the associated elements at the termination of the line, to match the the next available clock pulse from the pulse generator, characteristic impedance of the line as a whole, being, in they may be deliberately delayed still further by the use the typical example selected, 1600 ohms. The incoming of an arti?cial delay line of suitable electrical length so pulse (of approximately one-half microsecond duration that the pulse will arrive at the desired point in the correct synchronization. In other cases delay lines are used to 45 and 50 percent duty factor in SEAC) is delayed approxi mately 0.64 microsecond in the example shown, this being insure the synchronous arrival at a common point of required for the particular circuit shown in order to in pulses from different parts of SEAC which have fallen sure synchronization of the pulses from source 1 with out of synchronization owing to the characteristics of the other pulses from other sources. circuits through which they have passed. For these and The action of the circuit shown in Figure 1 will now other reasons, it is necessary to use electric delay lines in be described brie?y in order to show the manner of many places in the computer. application of the invention to the SEAC, since this is The actual electric switching operations in SEAC are a typical SEAC gate structure. Normally the grid of performed by germanium diodes. These or any similar tube 3 is maintained at —5 volts and there is no signal types of diodes are nonlinear devices; that is, their resist ance is not constant but is a function of the amplitude 55 output from the tube. To produce a signal it is desired that the grid be raised to +2 volts at a rapid rate (in and direction of the applied voltage. These diodes, to practice, the rate of 100 volts per microsecond), kept gether with their associated circuit elements, present a close to +2 volts for somewhat less than one-half micro nonlinear load termination to the delay lines where these second and dropped to —5 volts at the same rate of are used, and it has been found that such termination 100 volts per microsecond. This produces an ampli?ed presents a serious re?ection problem in some cases, as 60 output from tube 3 which results in a pulse output from will be shown more fully below. The production of a the secondary of transformer 4 similar to the original re?ected pulse travelling back along the delay line may input pulse and with at least the same power as the adversely affect some subsequent pulses in the same cir original input pulse from transformer 1. To provide cnit, since the re?ected pulse may arrive at the pulse source in the correct phase to in effect cancel out a subse the necessary safety factor, the voltage input from trans by simple readily available means. grid is kept at —5 volts by “bumper” 6, which is a circuit arrangement whereby voltage pulses of more than of the grid to drop below -5 results in current ?ow quent pulse and thus produce an error in the computation. 65 former 1 is made considerably higher than +2 volts. However, the input to the grid of tube 3 is kept to a This has in fact been found to occur in some parts of maximum of +2 volts by the action of the circuit which SEAC. It is a primary object of this invention to elimi will now be described. The potential of the unpulsed nate the possibility of errors occurring from this cause More speci?cally, it is a further object to provide a 70 diode connected to a —-5 volt source. Any tendency 2,727,143 3 4 through diode 6 to keep the grid up to at least —5 volts. If the grid potential rises above —-5 volts the anode is considerably higher, and current will ?ow as diode has no effect, as its cathode is now higher in potential than its anode and no current can ?ow through diode 8 ceases to conduct (at time T3) from which point the increase in current corresponds only to the increase in E1, i. e., El/Zo, which is a much slower rate of in crease than during the interval T2 to T3. During the time interval T2 to T3 the re?ected voltage Er is dropping as E1 is rising, since this is characteristic of short-circuited in a short circuit until it reaches the level at which it. ' Resistor 7 is Connected to a -—65 volt source and tends to pull the grid voltage down, but because of the above action the grid is maintained at —5 volts. Another diode, 8, is connected to a —-8 volt source. behavior. At time T3, the anode again appears more or Between diode 8 and resistor 7 is diode 9, which nor mally does not conduct because its anode is at a lower 10 less like an open circuit, i. e., the impedance now presented potential (—8 volts) than its cathode (—5 volts). Re to the pulse wave front is the relatively high impedance of sistor 11 is connected to a +62 volt source. resistors 11 and 13 in parallel. So'far as the gate is con cerned, soon after T3 the necessary voltage level for Current normally ?ows from resistor 11 and diode 8 to resistor 13 which is returned to —-65 volts. As 13 is of lower its proper operation is reached. However, the pulse volt resistance than 11, it tends to pull down the voltage 15 age continues to rise to its full value which is almost always higher than the minimum required for proper at diode 42 to below —8 volts. The cathode of diode gate operation. This excess voltage is the major cause 42 is, however, kept at ~8 volts by current supplied of re?ection trouble, because now E0 rises at twice the through diode 8. The source (secondary of transformer rate of Bi as E1- again builds up. The re?ected positive 1) is returned to —l0 volts so that normally no cur rent ?ows through diode 14, since its cathode is now 20 voltage wave Er may, in returning to its point of origin through the line, arrive in the correct phase to cause raised above its anode. However, an incoming pulse trouble. The characteristics of the gate circuit are such raises the potential of the anode of diode 14 to above that negative re?ections cause little di?iculty, except that greater power from the tube of the preceding stage will be required to overcome them in impressing subsequent pulses on the line, and there is usually su?icient power output available. Furthermore, examination of Figure 2a will show that the re?ected negative voltage pulse Er current now ?ows from resistor 11 through diode 9 and tends to be of smaller magnitude than the positive pulse, resistor 7. However, the rise is limited to +2 volts by diode 16 which is connected to a +2 volt source and 30 since it can occur only during the time interval T2 to T3 and so does not have much time to build up to a high begins to draw current to keep the grid at the +2 volt level whenever the grid voltage tends to rise above that negative value, while the positive pulse may build up to a fairly high value if the excess pulse voltage is suf point. Thus the grid voltage can ?uctuate only between ?ciently high. —5 volts and +2 volts regardless of the voltage of the incoming pulse. The above-described circuit is typical In a practical computer such as SEAC, the actual cir cuits tend to be rather complex, and the action is not of the gate circuitry of SEAC and is per se'no part always easy to follow. Perfect accuracy of pulse trans of the present invention. Similar circuits working on mission is, of course, required; i. e., not a single pulse the same principle are shown and described in the co may be lost or improperly transmitted without entirely pending claims of Ralph I. Slutz, Serial No. 193,696, ?led November 2, 1950, of William L. Martin et 2.1., 40 invalidating the computation, and with pulses at the rate of a million per second, in a run of some hours duration, Serial No. 205,l64, ?led January 9, 1951 and of Robert which often occurs, the number of possibilities of error D. Elbourn and Ralph J. Slutz, Serial No. 244,446, ?led is tremendous, yet the circuitry must be such that even August 30, 1951, now Patent No. 2,712,065, issued June the most remotely improbable combination of random 28, 1955. or transient effects is rendered innocuous. The above In order to explain the problem with whose solution described re?ection effect does not in practice give diffi this invention is concerned, the action of a part only +2 volts, and the resulting conduction raises the cathode of diode 42 above its anode potential. Since current can no longer ?ow from resistor 11 through resistor 13, the anode of 42 rises toward +62 volts. Diode 8 ceases to conduct because its cathode rises above -—8 volts and culty often, nor is it possible in some of the circuit com of the circuit shown in Figure 1 will now be considered. binations used, but the circuit combinations themselves Referring to Figures 2 and 2a, the characteristic im are electronically controlled by certain pulses acting as pedance at the line will be considered in its relation to the incident pulse front, and may be expressed as 50 coded orders rather than as numbers, and the circuits Z0 are therefore changed and rearranged at the same high speeds as those at which the computations are performed. E.;_ E, ‘If Under these circumstances it was found that errors can occasionally occur, and the provision of a way to success where E-;=incident pulse voltage lir=iricident pulse‘ current E}="re?ected ‘pulse voltage lr=re?ected pulse current Furthermore, E0==Ei+Er 55 fully correct the dif?culty was attained by the means de scribed below. where Eo=voltage at anode of diode 14. When the pulse front arrives at the anode of diode 14 (Figure 2), this point at ?rst behaves like an open circuit, since the diode prevents passage of current until 65 its anode voltage reaches the unlatching level of the diode, ' According to the present invention the possibility of delayed re?ections through lines of su?icient electrical length to produce harmful phase relationships is mini‘ 60 mized, or in practice totally eliminated, by providing an additional terminal load of such character that it does not appreciably affect the pulse action in the useful volt age range of the pulse, but does provide, in all higher voltage ranges, an apparent terminal impedance at least equal to, or below, the characteristic impedance of the line. At the characteristic impedance there is, of course, no re?ection, and at‘ a discontinuity representing a lower i. e., the level at which the diode begins to conduct. impedance than the characteristic impedance, the re?ected Therefore, from time T1 to T2 (Figure 2a) E0 r'ises'at voltage pulse is opposite in sign from the incident pulse twice the rate of E1 because of the addition of Er, as in any open-ended line re?ecting a wave front. After 70 and therefore does no harm, as above pointed. diode 14 begins to conduct, the line at this point behaves This new termination is accomplished by connecting like a short circuit from time T2 until the difference in the current in resistor 13 and resistor 11 is furnished entirely by the source, as at time T3. This is so be to the load end of the line, as shown in Figure 3b, a diode 21 returned to a voltage source of a value in the order of the maximum grid voltage—in this case +2 cause‘ the cathode of diode 14 is at -—8 volts and the 75 volts. In practice this return voltage may be slightly 2,727,143 higher, for example +4 volts, to provide a safety factor of voltage to take care of voltage drops further along the line toward the grid. The curves shown in Figure 4 show the effective impedance changes which occur in the corresponding circuits of Figure 3 as the voltage in 6 nonlinear impedance element whose impedance changes abruptly at approximately said de?nite voltage level said shunt being returned to a vintage source of a value sub creases. The dotted line Z0 represents the characteristic impedance of the line. Because of the nonlinear load stantially the value of said de?nite voltage level. 2. In an electronic system for transmitting and utiliz ing discrete electric pulses of very short duration at a very high repetition rate which comprises means for ing of the line by the gate element described, the im producing such pulses, a delay line for retarding pulses pedance in the range —l0 volts to +2 volts varies in by a predetermined amount, a nonlinear pulse utilization nonlinear fashion as shown. The curve shown in Figure 10 circuit connected to said delay line, said circuit having 4a represents the action in the circuit of Figures 1 and 2. means presenting an impedance not higher than the char At voltages above the maximum grid voltage of +2 volts acteristic impedance of said line to voltages of a certain the effective impedance is higher than the characteristic utilized value and a much higher impedance to voltages impedance, and therefore an undesirable positive re?ected in excess of said value whereby harmful re?ections may voltage occurs. Figure 4b shows impedance conditions occur because of said excess voltage value, the improve for the circuit of Figure 312. At the +2-volt grid level ment which comprises an additional nonlinear load for the load impedance drops suddenly because of the shunt ing action of diode 21 at voltages above this level, and thereafter is at all times less than the characteristic im pedance of the line, so that the undesirable positive pulse 20 re?ection does not occur. An alternative arrangement is shown in Figure 3c, which is very similar to Figure 3b except that a resist~ ance R0 is in series with the diode 21’. This resistor is preferably of such value that the parallel combination of it and the resistor 13 to the —65-volt supply is ap proximately equal to the characteristic impedance of the line. The corresponding impedance curve is shown in Figure 4c. This arrangement nicely absorbs the excess voltage from a single source, but is not so effective in the case of a gate having two inputs as shown in Figure 3d, if input A produces pulses of considerably greater amplitude than input B, in which case a large re?ection will be sent back toward source B if two pulses from both sources arrive simultaneously, with the undesirable effect previously noted. The apparent impedance to a pulse coming in from a source B under these conditions is represented in Figure 4d. It will be seen that the com posite impedance characteristic beyond +2 volts is higher than the characteristic impedance Z0 of the line. The arrangement of Figure 30 cannot therefore be used under these conditions, but should be used only where there is a single gate of the type shown in Figure 30. For the dual input of Figure 3d the arrangement of Figure 3b should be used, i. e., a diode returned to +2 volts without the resistor R0. By the provision of an additional nonlinear load of the type described, it has proved possible in SEAC to entirely eliminate harmful re?ection effects. SEAC, in which this invention is employed, has been able to operate on lengthy problems, involving many hours of continuous running, at a pulse rate of a megacycle, without a single erorr in pulse transmission. The present invention has been found to be a complete solution to the troublesome said system connected to a voltage source of said utilized value and so oriented as to present a high impedance to said pulses at voltages below said utilization value but a low impedance to said pulses at voltages above said utilization value, whereby said excess voltage is shunted and prevented from producing said harmful re?ections. 3. The invention as recited in claim 2 and a resistance in series with said additional nonlinear load, said re sistance being of such value that the total impedance presented to said pulses at voltages above said utiliza tion value is in the region of the characteristic impedance of the line. , 4. In a gating circuit for an electronic digital computer, said gating circuit including a pulse transformer having a primary and a secondary winding returned to a low reference voltage, a delay line connected to the high side of said secondary winding for retarding pulses pro duced in said secondary by a predetermined amount, a nonlinear pulse combining and amplifying circuit con nected to said delay line, said last circuit including re sistors and diode elements oriented and arranged to effectively transmit said pulses within a predetermined voltage range, said last circuit thereby presenting an impedance not higher than the characteristic impedance of said line to voltage pulses within said predetermined voltage range, and a higher impedance to voltages in excess of said value whereby such voltages may produce harmful re?ections, means for producing in said sec ondary, pulses of higher voltage than the upper level of said predetermined voltage range usable by said pulse combining and amplifying circuit, and means for pre venting harmful re?ections due to such excess voltage, said last means comprising an additional nonlinear ele ment at the utilization end of said delay line, said non linear element connected to a voltage source substan tially at the upper level of said predetermined voltage range, and being so oriented as to present a high im pedance to said pulses at voltages below said level but problem of harmful reflections despite the complex action 55 a low impedance to said pulses at voltages above said of the normal nonlinear load coupled to a delay line. utilization value whereby said excess voltage is shunted It will be apparent that the embodiments shown are and prevented from producing said harmful re?ections. only exemplary and that various modi?cations can be References Cited in the ?le of this patent made within the scope of my invention as de?ned in the appended claims. UNITED STATES PATENTS 60 Iclaim: 1. In a pulse utilization system comprising a delay line loaded with nonlinear pulse utilization means adapted for operation at a de?nite voltage level, means for prevent ing harmful re?ections which comprises a shunt con 65 nected across the load end of said delay line between said delay line and said load, said shunt comprising a 2,085,418 2,104,336 2,262,468 2,438,367 2,525,454 Crosby ______________ __ June 29, Tuttle _______________ __ Jan. 4, Percival _____________ .__ Nov. 11, Keister _____________ __ Mar. 23, Lord ________________ __ Oct. 10, 1937 1938 1941 1948 1950 2,557,122 Leiphart ____________ __ June 19, 1951